Commit 2ec08b61 authored by Uwe Schulzweida's avatar Uwe Schulzweida
Browse files

remapeta +OpenMP

parent ed290cf5
......@@ -4,6 +4,10 @@
#define OPOINT 34179
*/
#if defined (_OPENMP)
# include <omp.h>
#endif
#include <math.h>
#include <stdlib.h>
#include <stdio.h>
......@@ -72,6 +76,9 @@ static double esat(double temperature)
return es;
}
#define MAX_VARS 512
/* Source from INTERA */
void hetaeta(int ltq, int ngp, int *imiss,
int nlev1, double *ah1, double *bh1,
......@@ -86,7 +93,7 @@ void hetaeta(int ltq, int ngp, int *imiss,
static char func[] = "hetaeta";
double epsm1i, zdff, zdffl, ztv, zb, zbb, zc, zps;
double zsump, zsumpp, zsumt, zsumtp;
double dfi, fiadj= 0, dteta = 0;
double dfi, fiadj = 0, dteta = 0;
double pbl_lim, pbl_lim_need;
int jblt, jjblt;
int k, iv, ij, ijk, ijk1, ijk2;
......@@ -99,7 +106,6 @@ void hetaeta(int ltq, int ngp, int *imiss,
double *zvar;
double *rh_pbl = NULL;
double *theta_pbl = NULL;
double **vars_pbl = NULL;
double *rh2;
double *w1, *w2;
double *wgt;
......@@ -109,6 +115,19 @@ void hetaeta(int ltq, int ngp, int *imiss,
int nlev2p1;
int lpsmod = 1;
int klo;
#if defined (_OPENMP)
double **ph1_2, **lnph1_2, **fi1_2, **pf1_2, **lnpf1_2, **tv1_2, **theta1_2, **rh1_2, **zvar_2;
double **ph2_2, **lnph2_2, **fi2_2, **pf2_2;
double **rh_pbl_2, **theta_pbl_2, **rh2_2;
double **wgt_2;
double ***vars_pbl_2;
int **idx_2;
int ompthID, i;
extern int ompNumThreads;
double *vars_pbl[MAX_VARS];
#else
double **vars_pbl = NULL;
#endif
#if defined (OUTPUT)
double t, q, fi;
......@@ -119,14 +138,84 @@ void hetaeta(int ltq, int ngp, int *imiss,
nlev1p1 = nlev1+1;
nlev2p1 = nlev2+1;
#if defined (_OPENMP)
ph1_2 = (double **) malloc(ompNumThreads*sizeof(double *));
lnph1_2 = (double **) malloc(ompNumThreads*sizeof(double *));
fi1_2 = (double **) malloc(ompNumThreads*sizeof(double *));
pf1_2 = (double **) malloc(ompNumThreads*sizeof(double *));
lnpf1_2 = (double **) malloc(ompNumThreads*sizeof(double *));
tv1_2 = (double **) malloc(ompNumThreads*sizeof(double *));
theta1_2 = (double **) malloc(ompNumThreads*sizeof(double *));
rh1_2 = (double **) malloc(ompNumThreads*sizeof(double *));
zvar_2 = (double **) malloc(ompNumThreads*sizeof(double *));
ph2_2 = (double **) malloc(ompNumThreads*sizeof(double *));
lnph2_2 = (double **) malloc(ompNumThreads*sizeof(double *));
fi2_2 = (double **) malloc(ompNumThreads*sizeof(double *));
pf2_2 = (double **) malloc(ompNumThreads*sizeof(double *));
rh2_2 = (double **) malloc(ompNumThreads*sizeof(double *));
wgt_2 = (double **) malloc(ompNumThreads*sizeof(double *));
idx_2 = (int **) malloc(ompNumThreads*sizeof(int *));
if ( ltq )
{
rh_pbl_2 = (double **) malloc(ompNumThreads*sizeof(double *));
theta_pbl_2 = (double **) malloc(ompNumThreads*sizeof(double *));
}
if ( nvars > 0 )
{
vars_pbl_2 = (double ***) malloc(ompNumThreads*sizeof(double **));
}
for ( i = 0; i < ompNumThreads; i++ )
{
ph1_2[i] = (double *) malloc(nlev1p1*sizeof(double));
lnph1_2[i] = (double *) malloc(nlev1p1*sizeof(double));
fi1_2[i] = (double *) malloc(nlev1p1*sizeof(double));
pf1_2[i] = (double *) malloc(nlev1*sizeof(double));
lnpf1_2[i] = (double *) malloc(nlev1*sizeof(double));
tv1_2[i] = (double *) malloc(nlev1*sizeof(double));
theta1_2[i] = (double *) malloc(nlev1*sizeof(double));
rh1_2[i] = (double *) malloc(nlev1*sizeof(double));
zvar_2[i] = (double *) malloc(nlev1*sizeof(double));
ph2_2[i] = (double *) malloc(nlev2p1*sizeof(double));
lnph2_2[i] = (double *) malloc(nlev2p1*sizeof(double));
fi2_2[i] = (double *) malloc(nlev2p1*sizeof(double));
pf2_2[i] = (double *) malloc(nlev2*sizeof(double));
rh2_2[i] = (double *) malloc(nlev2*sizeof(double));
wgt_2[i] = (double *) malloc(nlev2*sizeof(double));
idx_2[i] = (int *) malloc(nlev2*sizeof(int));
if ( ltq )
{
rh_pbl_2[i] = (double *) malloc(nlev2*sizeof(double));
theta_pbl_2[i] = (double *) malloc(nlev2*sizeof(double));
}
if ( nvars > 0 )
{
if ( nvars > MAX_VARS )
{
fprintf(stderr, "To many vars (max = %d)!\n", MAX_VARS);
exit(-1);
}
vars_pbl_2[i] = (double **) malloc(nvars*sizeof(double *));
for ( iv = 0; iv < nvars; ++iv )
vars_pbl_2[i][iv] = (double *) malloc(nlev2*sizeof(double));
}
}
#else
/* etah1 = (double *) malloc(nlev1p1*sizeof(double)); */
ph1 = (double *) malloc(nlev1p1*sizeof(double));
lnph1 = (double *) malloc(nlev1p1*sizeof(double));
fi1 = (double *) malloc(nlev1p1*sizeof(double));
af1 = (double *) malloc(nlev1*sizeof(double));
bf1 = (double *) malloc(nlev1*sizeof(double));
etaf1 = (double *) malloc(nlev1*sizeof(double));
pf1 = (double *) malloc(nlev1*sizeof(double));
lnpf1 = (double *) malloc(nlev1*sizeof(double));
tv1 = (double *) malloc(nlev1*sizeof(double));
......@@ -134,16 +223,15 @@ void hetaeta(int ltq, int ngp, int *imiss,
rh1 = (double *) malloc(nlev1*sizeof(double));
zvar = (double *) malloc(nlev1*sizeof(double));
etah2 = (double *) malloc(nlev2p1*sizeof(double));
ph2 = (double *) malloc(nlev2p1*sizeof(double));
lnph2 = (double *) malloc(nlev2p1*sizeof(double));
fi2 = (double *) malloc(nlev2p1*sizeof(double));
af2 = (double *) malloc(nlev2*sizeof(double));
bf2 = (double *) malloc(nlev2*sizeof(double));
etaf2 = (double *) malloc(nlev2*sizeof(double));
pf2 = (double *) malloc(nlev2*sizeof(double));
/* lnpf2 = (double *) malloc(nlev2*sizeof(double)); */
rh2 = (double *) malloc(nlev2*sizeof(double));
wgt = (double *) malloc(nlev2*sizeof(double));
idx = (int *) malloc(nlev2*sizeof(int));
if ( ltq )
{
......@@ -151,22 +239,29 @@ void hetaeta(int ltq, int ngp, int *imiss,
theta_pbl = (double *) malloc(nlev2*sizeof(double));
}
wgt = (double *) malloc(nlev2*sizeof(double));
idx = (int *) malloc(nlev2*sizeof(int));
if ( nvars > 0 )
{
vars_pbl = (double **) malloc(nvars*sizeof(double *));
for ( iv = 0; iv < nvars; ++iv )
vars_pbl[iv] = (double *) malloc(nlev2*sizeof(double));
}
rh2 = (double *) malloc(nlev2*sizeof(double));
w1 = (double *) malloc(nlev2*sizeof(double));
w2 = (double *) malloc(nlev2*sizeof(double));
jl1 = (int *) malloc(nlev2*sizeof(int));
jl2 = (int *) malloc(nlev2*sizeof(int));
#endif
af1 = (double *) malloc(nlev1*sizeof(double));
bf1 = (double *) malloc(nlev1*sizeof(double));
etaf1 = (double *) malloc(nlev1*sizeof(double));
etah2 = (double *) malloc(nlev2p1*sizeof(double));
af2 = (double *) malloc(nlev2*sizeof(double));
bf2 = (double *) malloc(nlev2*sizeof(double));
etaf2 = (double *) malloc(nlev2*sizeof(double));
w1 = (double *) malloc(nlev2*sizeof(double));
w2 = (double *) malloc(nlev2*sizeof(double));
jl1 = (int *) malloc(nlev2*sizeof(int));
jl2 = (int *) malloc(nlev2*sizeof(int));
/******* set coordinate system ETA's, A's, B's
calculate half and full level ETA
......@@ -232,8 +327,7 @@ void hetaeta(int ltq, int ngp, int *imiss,
if (etaf2[k] > etaf1[jlev]) break;
}
jl2[k] = jl1[k]+1;
w2[k] = log(etaf2[k]/etaf1[jl1[k]])
/log(etaf1[jl2[k]]/etaf1[jl1[k]]);
w2[k] = log(etaf2[k]/etaf1[jl1[k]]) / log(etaf1[jl2[k]]/etaf1[jl1[k]]);
}
w1[k] = 1.0-w2[k];
}
......@@ -241,8 +335,56 @@ void hetaeta(int ltq, int ngp, int *imiss,
epsm1i = 1.0/epsilon-1.0;
#if defined (_OPENMP)
#pragma omp parallel for default(none) \
shared(ngp, ph1_2, lnph1_2, fi1_2, pf1_2, lnpf1_2, tv1_2, theta1_2, rh1_2, zvar_2, ph2_2, lnph2_2, fi2_2, pf2_2, rh_pbl_2, \
theta_pbl_2, rh2_2, wgt_2, vars_pbl_2, idx_2, af1, bf1, etaf1, etah2, af2, bf2, etaf2, w1, w2, jl1, jl2, \
ltq, nvars, imiss, nlev1p1, ah1, bh1, ps1, nlev1, epsm1i, q1, t1, fis1, fis2, ps2, nlev2p1, ah2, bh2, \
nlev2, vars1, vars2, t2, q2, tscor, pscor, secor, jblt, stderr) \
firstprivate(lpsmod) \
private(ij, iv, ph1, lnph1, fi1, pf1, lnpf1, tv1, theta1, rh1, zvar, ph2, lnph2, fi2, pf2, rh_pbl, theta_pbl, rh2, wgt, vars_pbl, \
k, ijk, jlev, idx, ompthID, zdff, jlevr, zdffl, jnop, zsumt, zsump, zsumpp, zsumtp, zb, zc, zps, zbb, \
fiadj, pbl_lim, jjblt, pbl_lim_need, ijk1, ijk2, klo, dteta, dfi, ztv) \
schedule(dynamic,1)
#endif
for ( ij = 0; ij < ngp; ++ij )
{
#if defined (_OPENMP)
ompthID = omp_get_thread_num();
ph1 = ph1_2[ompthID];
lnph1 = lnph1_2[ompthID];
fi1 = fi1_2[ompthID];
pf1 = pf1_2[ompthID];
lnpf1 = lnpf1_2[ompthID];
tv1 = tv1_2[ompthID];
theta1 = theta1_2[ompthID];
rh1 = rh1_2[ompthID];
zvar = zvar_2[ompthID];
ph2 = ph2_2[ompthID];
lnph2 = lnph2_2[ompthID];
fi2 = fi2_2[ompthID];
pf2 = pf2_2[ompthID];
rh2 = rh2_2[ompthID];
wgt = wgt_2[ompthID];
idx = idx_2[ompthID];
if ( ltq )
{
rh_pbl = rh_pbl_2[ompthID];
theta_pbl = theta_pbl_2[ompthID];
}
if ( nvars > 0 )
{
for ( iv = 0; iv < nvars; ++iv )
vars_pbl[iv] = vars_pbl_2[ompthID][iv];
}
#endif
if ( imiss )
if ( imiss[ij] ) continue;
......@@ -587,59 +729,128 @@ void hetaeta(int ltq, int ngp, int *imiss,
fclose(new);
#endif
free(jl2);
free(jl1);
free(w2);
free(w1);
#if defined (_OPENMP)
for ( i = 0; i < ompNumThreads; i++ )
{
free(ph1_2[i]);
free(lnph1_2[i]);
free(fi1_2[i]);
free(pf1_2[i]);
free(lnpf1_2[i]);
free(tv1_2[i]);
free(theta1_2[i]);
free(rh1_2[i]);
free(zvar_2[i]);
free(ph2_2[i]);
free(lnph2_2[i]);
free(fi2_2[i]);
free(pf2_2[i]);
/* free(lnpf2_2[i]); */
free(rh2_2[i]);
free(wgt_2[i]);
free(idx_2[i]);
free(rh2);
if ( ltq )
{
free(rh_pbl_2[i]);
free(theta_pbl_2[i]);
}
if ( nvars > 0 )
if ( nvars > 0 )
{
for ( iv = 0; iv < nvars; ++iv )
free(vars_pbl_2[i][iv]);
free(vars_pbl_2[i]);
}
}
free(ph1_2);
free(lnph1_2);
free(fi1_2);
free(pf1_2);
free(lnpf1_2);
free(tv1_2);
free(theta1_2);
free(rh1_2);
free(zvar_2);
free(ph2_2);
free(lnph2_2);
free(fi2_2);
free(pf2_2);
free(rh2_2);
free(wgt_2);
free(idx_2);
if ( ltq )
{
for ( iv = 0; iv < nvars; ++iv )
free(vars_pbl[iv]);
free(rh_pbl_2);
free(theta_pbl_2);
}
free(vars_pbl);
if ( nvars > 0 )
{
free(vars_pbl_2);
}
#else
/* free(etah1); */
free(ph1);
free(lnph1);
free(fi1);
free(idx);
free(pf1);
free(lnpf1);
free(tv1);
free(theta1);
free(rh1);
free(zvar);
free(ph2);
free(lnph2);
free(fi2);
free(pf2);
/* free(lnpf2); */
free(rh2);
free(wgt);
free(idx);
if ( ltq )
{
free(theta_pbl);
free(rh_pbl);
free(theta_pbl);
}
/* free(lnpf2); */
free(pf2);
free(etaf2);
free(bf2);
free(af2);
free(fi2);
free(lnph2);
free(ph2);
free(etah2);
if ( nvars > 0 )
{
for ( iv = 0; iv < nvars; ++iv )
free(vars_pbl[iv]);
free(zvar);
free(rh1);
free(theta1);
free(tv1);
free(vars_pbl);
}
#endif
free(lnpf1);
free(pf1);
free(af1);
free(bf1);
free(etaf1);
free(bf1);
free(af1);
free(etah2);
free(fi1);
free(lnph1);
free(ph1);
/* free(etah1); */
free(af2);
free(bf2);
free(etaf2);
free(w1);
free(w2);
free(jl1);
free(jl2);
return;
}
......
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment